分布式光伏发电特性与气象影响因子诊断分析

Diagnostic Analysis of Distributed Photovoltaic Power Characteristics and the Impact of Meteorological Factors

为了实现光伏电站并入电网后能够安全稳定运行以及农村电力系统发电计划的制定,光伏电站发电功率的准确预测是必不可少的。通过采集沈阳地区分布式光伏电站2014年10月至2016年9月发电功率与气象现场测试数据,利用Pearson相关性分析方法对光伏发电功率与同期气象影响因子进行了相关性分析:太阳辐射量、日照时数和日最高气温与光伏综合出力相关性最高,相关系数分别为0.902,0.782,0.364;在此基础上分析这3种气象因子在不同季节下与光伏发电功率的相关程度:夏季太阳辐射量和日照时数与发电功率相关程度最高,分别为0.972和0.641,秋季日最高气温与发电发电功率相关程度最高,相关性为0.382。在不同季节的基础上分析了不同天气类型下(晴、阴/多云、多云/晴、阴雨、多云、晴/霾和雪/多云)发电功率的扰动程度:不同季节、不同天气类型下日发电功率曲线均呈现正态分布,其中晴天发电功率扰动最小、阴雨天气发电功率扰动最大,晴天、多云、多云/晴、阴云、阴雨天的四季平均标准偏差分别为1.44,2.81,3.12,3.36,3.51,晴/霾和雪天的标准偏差均为1.91。将太阳辐射量、日照时数和日最高气温作为输入,建立不同季节不同天气类型发电功率多元线性回归模型,对2016年10月发电功率进行预测,试验结果表明:预测误差均小于20%,满足电网要求,发电功率的准确预测可以更好地实现农村电网的管理和调度。

In order to realize the safe and stable operation of the rural power grid after the PV power plant is integrated into the grid and the formulation of the power generation plan for the rural power system, accurate prediction of the power generated by the photovoltaic power plant is indispensable. Through the acquisition of power generation and meteorological field test data of distributed photovoltaic power stations in Shenyang from October 2014 to September 2016, the Pearson correlation analysis method was used to analyze the correlation between photovoltaic power generation and meteorological factors in the same period.The correlation between sunshine hours and daily maximum temperature and PV integrated output was the highest, and the correlation coefficients were 0.902, 0.782 and 0.364, respectively. On this basis, the degree of correlation between the three meteorological factors and photovoltaic power generation in different seasons was analyzed. The correlations between radiation and sunshine duration and power generation was the highest, which were 0.972 and 0.641, respectively. The highest correlation between the maximum daily temperature and generation power in autumn was 0.382. Based on different seasons, the degree of disturbance of power generation under different weather types（sunny, cloudy/partly cloudy, partly cloudy/sunny, rainy, partly cloudy, sunny/fog and snow/partly cloudy） was analyzed, and daily power generation was conducted under different seasons and weather types. The power curves showed a normal distribution. Among them, the minimum power generation disturbance occurred in sunny days and the largest power disturbance occurred in rainy days, and the average standard deviations for the four seasons were 1.40, 2.81, 3.12, 3.36, and 3.51 for sunny, partly cloudy, partly cloudy/sunny, cloudy, and rainy days, respectively. The standard deviation of both sunny/fog and snow days was 1.91. With solar radiation, sunshine hours, and daily maximum temperature as inputs, a multivariate linear regression model for different types of weather generation in different seasons was established to predict the power generation in October 2016. The test results showed that the prediction errors were all less than 20%, and the grid power requirements were accurately met. Forecasting can better achieve the management and dispatch of rural power grids.